Blog
Insights on AI agents, web accessibility, and getting your site ready for the agent era.
The 32-point gap
Every site loaded for the agent. Not every agent got the job done. 185 real audits across 94 unique sites, 13 Jan to 21 Apr 2026.
Children's Activity Centres: Agent Comparison Shopping for Parents
Activity centres with structured timetable data, age ranges and pricing in HTML are easy for AI agents to recommend. Those relying on PDF timetables tend to be skipped.
Home Energy Assessors and Agent-Booked Appointments
The postcode-to-address-to-appointment booking pattern is almost impossible for AI agents. URL-based steps, native controls and clear postcode handling make it work.
Event Ticketing: When Agents Buy Seats on Behalf of Fans
Interactive seat maps are an impenetrable barrier for AI agents. A list-based seat selection alternative alongside the map lets agents buy seats fans actually want.
What the Next Generation of Web Agents Will Look Like
Based on current research and development trends, here is what web-browsing AI agents will likely be able to do in the near future.
Accessibility Overlays: Do They Help or Hinder AI Agents
Third-party accessibility overlays promise automatic WCAG compliance by injecting JavaScript. But the DOM changes they introduce often confuse AI agents more than they help.
How Agents Handle Payment Flows and Checkout Security
Payments are the hardest step for autonomous agents. 3DS challenges, tokenised credentials, and emerging agent-specific payment patterns are reshaping checkout design.
Coworking Spaces: Availability Listings That Agents Understand
Interactive floor plans and canvas timelines are opaque to AI agents. A structured listing view, JSON-LD markup and an availability API let agents book desks and rooms.
How AI Agents Handle Authentication and Sessions
Authentication is one of the hardest problems for browser-based AI agents. Here is how they manage logins, sessions, and protected content.
Content Negotiation: Serving Machine-Readable Formats to Agents
Content negotiation lets your server return different formats based on what the client asks for. Agents that send Accept: application/json could get structured data from the same URL that serves HTML to browsers.
Search Filters That AI Agents Can Actually Operate
Native HTML checkboxes and select elements are easy for AI agents to operate, while custom JavaScript sliders, colour pickers, and drag-to-filter widgets are often unusable.
Comparing Browser Automation Frameworks: What Agents Use Under the Hood
A practical comparison of Playwright, Puppeteer, Selenium, and other tools that power browser-based AI agents.
Headless Browsers vs HTML Parsing: How Agents Choose
Some agents fetch raw HTML. Others spin up full browser instances. The choice depends on the site, and it affects which sites agents can actually work with.
Pharmacy Delivery: How Agents Check Stock and Place Orders
Colour-only stock badges leave AI agents unable to tell what is in stock. Text labels and structured Product data let agents check availability and reorder reliably.
Progressive enhancement for both humans and agents
Progressive enhancement is not just a front-end philosophy. It is the best strategy for building sites that work for humans, agents, and everything in between.
WebSocket Connections: Can Agents Use Them?
Most AI agents cannot establish WebSocket connections. Content and features that depend entirely on WebSockets are invisible to agents unless you provide HTTP fallbacks.
The ROI of an Agent Readiness Audit
An agent readiness audit costs a fraction of what you're losing by being invisible to AI agents. Here's how to calculate the return for your specific business.
The Rise of Agent-to-Agent Communication on the Web
AI agents are beginning to coordinate with each other through web infrastructure. Here is how that works and where it is heading.
How Modal Dialogs Block Agent Workflows
Newsletter popups, age gates, consent forms, and exit-intent modals all force AI agents to dismiss an unexpected overlay before they can access any page content.
Making your search function work for AI agents
Site search is one of the first things an AI agent tries. If your search is broken for agents, they will struggle with everything else too.
Server-Sent Events and Streaming Data for Agents
Server-Sent Events push real-time updates to browsers, but most agents cannot consume them. If your critical data only arrives via SSE, agents miss it entirely.
The Latency Budget: How Long Agents Wait Before Moving On
AI agents work under strict time constraints. If your page does not return useful data fast enough, the agent skips you entirely.
Wedding Venues: Why Availability Data Beats Photos for Agents
Beautiful wedding venue sites with no structured availability, capacity or pricing are invisible to AI agents. Machine-readable data is what wins agent-referred enquiries.
Agent Memory and Context: Why Repeat Visits Matter
AI agents are starting to remember what they learned on previous visits to your site. This changes how they interact with your pages over time.
How Small Businesses Can Compete for Agent Traffic
AI agents don't care about your marketing budget, your brand recognition, or your domain authority. They care about data quality. That's an advantage small businesses should be using.
Structured data testing: tools and approaches that work
You have added Schema.org markup. How do you know it actually works? Here are the tools and testing approaches that catch real problems.
Performance Budgets and Agent Timeout Thresholds
Agents abandon slow pages. If your site does not deliver content within a few seconds, agents move on. Here is how to align your performance budgets with agent expectations.
Multi-Language Sites and the Agent Locale Problem
When a site has multiple language versions, AI agents can land on the wrong locale, get redirected by geolocation, or miss content that only exists in one language.
How Large Language Models Interpret Web Page Layouts
What actually happens when you feed a web page to a language model, and why some pages are easier for AI to understand than others.
Why Agents Prefer APIs Over Web Interfaces
Given the choice between browsing a web page and calling an API, AI agents pick the API every time. Structured data, predictable responses, and no visual parsing required.
How Regional Airports Can Fix Agent-Driven Flight Search
Regional airport sites route passengers to airlines, but custom datepickers and scattered schedules block AI agents. Structured route data and native form controls fix it.
API-first design: giving agents a better interface than your website
Your website was designed for humans. An API gives AI agents a structured, efficient way to interact with your services without scraping HTML.
Hreflang Tags and How Agents Route International Traffic
Hreflang tags tell agents which language or regional version of a page to use. Errors in hreflang implementation leave agents guessing, often sending users to the wrong version.
How AI Agents Reduce Subscription Churn
Users are deploying agents to monitor their subscriptions, compare alternatives, and cancel bad deals. Subscription businesses that work with agents will keep more customers than those that hide behind dark patterns.
Tool-Calling Agents and Why Your Website Is Just Another Tool
AI agents increasingly treat websites as callable tools. Here is what that means for how they interact with your pages and APIs.
Pagination vs Load More: Which Pattern Works Better for Agents
Traditional pagination with numbered page links gives AI agents predictable URLs and full catalogue access, while Load More buttons and infinite scroll leave most content unreachable.
HTTP status codes and how they affect agent browsing
AI agents rely on HTTP status codes to make decisions about retries, redirects, and error handling. Wrong codes cause agents to behave in unexpected ways.
Canonical URLs and How Agents Handle Duplicate Content
Without canonical tags, agents may waste time crawling duplicate versions of your pages. Here is how canonical URLs help agents find the right version and skip the rest.
Retrieval-Augmented Browsing: How Agents Build Memory Across Sessions
AI agents are starting to remember previous visits to your site. What they store, how they retrieve it, and why keeping your content accurate matters more than ever.
Pet Insurance: What Agents Look for in Policy Comparison Pages
On pet insurance comparison pages, the details that matter are often hidden behind accordions agents cannot open. Visible HTML and structured data make comparison work.
Multi-Modal Agents: When AI Can See Your Website Like a Human
How vision-capable AI agents perceive web pages, and why layout and visual design now matter for machine audiences too.
Agent-Friendly Sites Convert Better for Humans Too
The improvements you make for AI agents, cleaner HTML, clearer structure, plain-language descriptions, also make your site better for human visitors. It's not an either/or trade-off.
The role of sitemaps and robots.txt in the age of AI agents
Sitemaps and robots.txt are not just for search crawlers. AI agents use them to understand your site structure and know where they are allowed to go.
Web Components and Shadow DOM: The Agent Compatibility Question
Shadow DOM hides internal structure from external queries. When agents try to read web components, they often see custom tags with no visible content inside.
Breadcrumbs That Help Agents Map Your Site Structure
Breadcrumb navigation gives AI agents a clear page hierarchy they can read without crawling your entire site, especially when paired with BreadcrumbList schema markup.
The Evolution from Web Scraping to Intelligent Browsing
How web data extraction went from regex on raw HTML to AI agents that understand pages the way people do.
Agent Orchestration: When Multiple Agents Tackle One Task
Complex tasks increasingly get split across specialised agents. A research agent visits your site first, then hands off findings to a purchasing agent.
How AI Agents Filter Job Listings on Recruitment Platforms
Custom JavaScript filters with no URL state stop AI agents from searching jobs. Native form elements and URL-based filter state make precise filtering possible.
Building agent-friendly forms: input types, labels, and validation
Forms are where AI agents do real work on your site. Poorly built forms are the number one reason agents fail to complete tasks.
Lazy Loading and What Agents Actually See
Lazy loading saves bandwidth for human visitors, but agents that do not scroll may never trigger it. Here is what agents actually see on lazy-loaded pages.
Brand Visibility When Customers Never Visit Your Site
If customers only interact with your brand through an agent's recommendations, your website design becomes irrelevant. Structured data becomes your brand's face.
Browser-Using Agents: How They Work and What They Expect
A practical look at how AI agents interact with websites through browser automation, and what your site needs to do to keep up.
Sticky Headers and Fixed Overlays: What Agents Cannot Click Past
Fixed-position headers, floating CTAs, and chat widgets sit on top of page content and intercept the clicks that browser-based AI agents are trying to make.
How to structure your HTML so AI agents can parse it
Clean, semantic HTML is the foundation of agent-friendly websites. Here is a practical guide to structuring your markup for both humans and machines.
Content Security Policies and Their Effect on Agent Browsing
Strict Content Security Policies can break pages when AI agents visit them in headless browsers. Here is how CSP interacts with agent browsing and what to do about it.
Vector Embeddings and How Agents Understand Page Content
Some AI agents convert your page content into vector embeddings for later retrieval. How well your content embeds depends on how well it is structured.
Charity Donation Pages That AI Agents Can Actually Complete
Hidden form controls and embedded payment iframes stop AI agents from donating. Native form controls, clear step progression and a redirect payment flow fix it.
When Agents Match Customers Better, Returns Drop
Agent-referred purchases have measurably lower return rates. Better matching means fewer disappointed customers, lower logistics costs, and healthier margins.
ARIA labels aren't just for screen readers anymore
ARIA attributes have become a primary way AI agents identify and interact with page elements. Here is what that means for your markup.
Rate Limiting AI Agents Without Blocking Them Entirely
Aggressive rate limiting can lock out legitimate AI agents. Too little and your servers get hammered. Here is how to find the right balance.
How Mega Menus Confuse AI Agent Navigation
Mega menus pack hundreds of navigation links behind hover interactions that AI agents cannot trigger, making most of your site structure invisible.
How Agents Decide Which Links to Follow
A typical web page has dozens of links. AI agents use a mix of text analysis, position, and context to decide which ones matter for their task.
Fitness Studios and AI Agents: Class Booking That Works
Visual calendar grids look great but block AI agents completely. Structured timetable data and accessible time slot buttons let agents book classes for members reliably.
Schema.org markup: the language AI agents actually understand
AI agents rely on structured data to make sense of your pages. Here is how to implement Schema.org markup that agents can actually work with.
JSON-LD vs Microdata: Which Format Agents Prefer
Agents need structured data to understand your pages, but the format you choose affects how easily they can extract it. JSON-LD and Microdata serve the same purpose with very different trade-offs.
Agent Referral Traffic Is a Channel You Cannot Ignore
You track organic, paid, and social traffic. But agent referral traffic is growing faster than any of them, and most analytics setups miss it entirely.
Tab-Based Layouts and Why Agents Miss Hidden Content
Tabbed interfaces hide most of a page behind inactive panels, leaving AI agents with only the default tab content and no awareness that the rest exists.
Computer Vision in Web Agents: When Text Is Not Enough
How modern AI agents use vision models to understand web pages beyond raw HTML, and what that means for sites that rely on visual communication.
Why Energy Supplier Booking Wizards Fail AI Agents
Smart meter and service booking wizards built for human eyes often block AI agents entirely. Semantic HTML, native form controls and URL-based steps make them work.
Open Graph and Meta Tags: What Agents Read Before They Visit
AI agents scan your meta tags before they even load the page. The right Open Graph and meta description setup determines whether an agent bothers to visit at all.
Why Your Pricing Page Loses Agent-Referred Customers
AI agents are sending qualified buyers to your pricing page, and your layout is driving them away. The patterns that confuse agents are fixable, but only if you know what to look for.
B2B SaaS: Why Your Pricing Page Confuses AI Agents
Animated price counters, icon-only feature grids, hidden monthly toggles, and "Contact us" enterprise tiers mean AI agents often cannot describe SaaS pricing accurately.
Grocery Delivery: How Agents Handle Product Catalogues
Grocery catalogues are hard for agents: inconsistent product names, allergen info trapped in images, no unit pricing, and session-only baskets with no shareable URL.
Automotive Dealerships: How AI Agents Compare Vehicles Online
Dealer websites often throw away the rich DMS feed data agents need: specs become unlabelled text, mileage lives in image overlays, and filters use invisible hash fragments.
Education Platforms and Agent-Driven Course Recommendations
Online learning catalogues often fail agents because duration is ambiguous, pricing is incomplete, and prerequisites are buried in prose. Restructuring for readability lifts enrolments.
First-Mover Advantage in Agent-Ready Web Design
The window for gaining early advantage in agent accessibility is open right now, but it won't stay open long. Here's what to do about it.
Legal Services and AI Agents: Making Complex Information Accessible
Law firm websites are often nearly opaque to agents: pricing buried in long prose, no structured data, and enquiry forms hidden behind modals. Restructuring for readability lifts enquiries.
Banking and Finance: When AI Agents Need to Read Your Product Pages
When interest rates are rendered as SVGs and key facts hide behind accordions, AI agents cannot accurately describe financial products, and may even quote the wrong rate.
Agent-Driven Commerce: The Shift from Search to Action
Online shopping is evolving from humans browsing and clicking to agents comparing, deciding, and purchasing. This changes what an e-commerce site needs to do.
Restaurant Discovery: How AI Agents Choose Where to Send Customers
When AI agents can actually read your menu, hours, and booking system, reservations from agent referrals can grow. PDF-only menus and image-based hours make a restaurant invisible to them.
Insurance Comparison Through AI Agents: What Works and What Doesn't
Insurance comparison sites range from impossible to easy for agents depending on whether premiums are real text, whether policy detail hides behind modals, and how forms are built.
Why Your Bounce Rate Might Actually Be an Agent Problem
That rising bounce rate might not be a content quality issue. It could be AI agents hitting your site, failing to parse it, and leaving immediately.
Travel Booking Sites That AI Agents Actually Prefer
Two travel booking sites can have similar inventory yet very different agent accessibility, depending on whether listings are real HTML or images, where CAPTCHAs sit, and how prices are assembled.
The Cost of Ignoring AI Agent Traffic
Ignoring AI agent traffic isn't a neutral decision. It actively damages your competitive position in ways that compound over time.
Estate Agents and AI: Why Property Listings Need Structured Data
Property listings are well-organised for human browsing but often leave key facts trapped in prose, CSS-class icons, and certificate images, so agents extract accurate details from very few of them.
Healthcare Portals and AI Agents: Booking Appointments Without Friction
Patient portals often let agent-initiated bookings fail silently because of div-based calendar grids, AJAX wizards with no URL changes, and custom checkboxes. Here is how to fix it.
Measuring Your Site's Agent Readiness: The Metrics That Matter
You can't improve what you can't measure. Here are the specific metrics that tell you how well your site works for AI agents, and how to track them.
How Home Furnishings Retailers Can Grow Agent-Driven Sales
Home furnishings retailers often look fine to shoppers but fail AI agents because of JavaScript prices, icon-only stock indicators, and missing structured data. Here is how to fix it.
Why AI Agents Abandon Complex Checkout Flows
Multi-step checkouts with address forms, payment selection, upsells, and confirmation pages are where AI agents fail most often, right when they are closest to completing a purchase.
The CAPTCHA Dilemma: Security vs Agent Accessibility
CAPTCHAs exist to block bots, and AI agents are technically bots. Solving this tension without compromising security is one of the hardest problems in agent accessibility.
AI Agents as the New Search Engines: What This Means for Traffic
The way people find and choose products is shifting from search-and-click to ask-and-delegate. Your traffic acquisition strategy needs to account for this.
How Client-Side Rendering Hides Content from AI Agents
Client-side rendered pages ship an empty HTML shell and rely entirely on JavaScript to build the content, which many AI agents never see.
How Agent-Ready Competitors Are Capturing Your Customers
While you debate whether AI agents matter, your competitors are quietly making their sites agent-friendly and picking up the customers those agents send their way.
Dynamic Content Loading and the Timing Problem for Agents
When page content loads in waves, AI agents have to guess when the page is "done." Guess too early and they miss data. Wait too long and they waste time on every page.
The Hidden Revenue You're Losing to Poor Agent Accessibility
Most businesses have no idea how much money walks out the door when AI agents can't interact with their websites. The numbers are bigger than you think.
Image-Heavy Sites and the Information Gap for AI Agents
When critical information exists only in images, banners, and infographics, AI agents that rely on text parsing miss most of what makes your site useful.
Why Login Walls Are Invisible to AI Shopping Agents
Requiring an account to view prices, stock levels, or product details means AI shopping agents hit a dead end where humans see a login form.
Cookie Consent Popups Blocking AI Agents from Content
Cookie consent banners can cover half the screen, trap keyboard focus, and prevent page interaction until dismissed, leaving AI agents stuck on a wall of legal text.
How Infinite Scroll Breaks AI Agent Browsing
Infinite scroll replaces pagination links with scroll-triggered content loading, which means AI agents see only the first batch of results and have no way to access the rest.
The Dropdown Menu Problem: Why Agents Can't Find Your Navigation
Hover-triggered dropdown menus are invisible to most AI agents because the menu items only exist in the DOM after a mouse event that agents never fire.
Why AI Agents Struggle with Single-Page Applications
Single-page applications load content without changing the URL, which leaves most AI agents stranded on a blank shell with no idea where the actual content lives.